parallel decoding
Parallel decoding generates several tokens in a sequence at the same time, rather than one after another. This speeds up generation by predicting multiple positions simultaneously, often through iterative refinement or blockwise coordination. It avoids the strict left-to-right bottleneck of traditional models.
The papers show that parallel decoding matches or exceeds autoregressive quality while being faster. They find it works with hybrid backbones, improves grounding tasks, and can be trained directly from sequence data without separate components. The speed and quality benefits grow with model size, making it a practical alternative to sequential generation.
- Papers
- 3
- Released code
- 0
- First seen
- Sept 2026
- Latest
- Oct 2026
2 papers in the last two months, against 1 in the two before.
Start here
Two papers to read first, in the order they came out- 1Hybrid backbones can efficiently adapt into diffusion language models
Hybrid-attention diffusion models from Qwen3.5 adapt faster and decode in any order and in parallel.
Sept 2026 - 2Parallel diffusion decoding rivals autoregressive visual grounding at 4B scale
GroundAnything uses blockwise diffusion to deliver faster parallel visual grounding while retaining strong localization accuracy across diverse tasks.
Oct 2026
How fast it is moving
Often studied with
speculative decodingA method that uses a lightweight model to propose token sequences that are then verified in parallel by a larger model.2 togetherdiffusion language modelA language model that generates text by iteratively denoising a latent or discrete representation.1 togetherdiffusion modelA type of generative model that learns to reverse a gradual noising process to create data.1 togetherThe papers
Most central to this idea first, not most recent.
- Independentcs.CV
Parallel diffusion decoding rivals autoregressive visual grounding at 4B scale
GroundAnything uses blockwise diffusion to deliver faster parallel visual grounding while retaining strong localization accuracy across diverse tasks.
Oct 2026
- Academiccs.CL
Hybrid backbones can efficiently adapt into diffusion language models
Hybrid-attention diffusion models from Qwen3.5 adapt faster and decode in any order and in parallel.
University of Texas at Austin · Sept 2026
- Top Universitycs.LG
Parallel language models match autoregressive quality while decoding several tokens per pass
Parallel Language Models generate coordinated tokens per forward pass at autoregressive-level quality, delivering 2.3-5.7× faster inference in reported post-trained models.
University of California, Irvine, ETH Zurich · Oct 2026
Concepts are extracted from each paper and reused across the corpus, so this page grows on its own as the desk reads.